2. Overstatements in USDA Census of Agriculture Data

USDA reports many commonly cited statistics about farms and farm income that are distorted by the way the department’s Census of Agriculture counts farms. For example, the Wall Street Journal reported in 2019 that “more than half of U.S.

Many of these problems relate to the Census of Agriculture’s methodology. Even as fewer farms came to control more and more acreage, the agricultural census began to register a sharp increase in farms after USDA took over the survey in 1997. At that time, the department introduced a series of methodological changes designed to increase the survey’s counts.¹⁴ By 2017, these changes had brought the total count of farms back to roughly the same level as 1987, even as other sources of data on farms continued to show a decline. The number of households filing Schedule F forms with the Internal Revenue Service (IRS), which are used to report farm income and expenses, declined by 34% between 1978 and 2017, and the number of farm households identified by the Current Population Survey (CPS), a federal survey conducted by the Census Bureau that is the source of national employment statistics, declined by 35%. Meanwhile, the agricultural census showed a decline of only 17% during this period. All three data sources showed similar trends until 1997. Today, the agriculture census shows twice as many “farms” as the CPS.¹⁵

USDA changed the agricultural census, at least in part, to address its historic undercounts of small-scale and non-white farmers—especially Black and indigenous farmers. But the department overcorrected and now counts a large number of non-farms as farms.

16

have been sold, during the census year.”¹⁶ Had USDA adjusted the $1,000 income threshold for inflation, that alone would have excluded almost half of all “farms” in the 2017 Census of Agriculture.¹⁷

Not only did USDA not adjust the income threshold, it also broadened its interpretation of “normally would have been sold.” The department devised a point system to estimate how much income a plot of land could produce if it were used to raise or grow agricultural products—even if the operator had never used the land as agricultural land and the owners had no intention of using it that way.¹⁸ Rural homes with berry bushes (at least one-fifth of an acre), vegetables (one-fifth an acre), horses (10 acres of pasture), cattle (one acre), or other potential “agricultural products” all count as farms under the official definition.¹⁹

Since 1997, the agricultural census has included a greater and greater share of “point farms”: properties that met the definition of “farm” because USDA estimated that they could have sold, but did not sell, $1,000 of agricultural products.²⁰ By 2017, almost 30% of farms in the agricultural census were point farms²¹ and more than 20% of census “farms” did not sell any agricultural products whatsoever.²² In fact, well over half of all farms reported in the agricultural census are, by the department’s own definitions,²³ not farm businesses, but retirement or “lifestyle” farms; this latter category was so-named “because many of the operators on these farms view their farms largely as an avocation or a place to live where they can enjoy a rural lifestyle.”²⁴ As one

17

journalist put it, most small farmers in the census “aren’t the farms of the poor; they’re the yards of the upper-middle-class.”²⁵

Another important, but harder to quantify source of overstatement of the number of farms comes from people who define their property as a farm for tax purposes. Farm operations receive numerous tax benefits, notably lower property taxes, which encourage property owners to classify their land as “agricultural.” All 50 states offer “use-value assessment” for agricultural land, which allows owners to assess their property at rates well below market value, often by 90% or more.²⁶ Many states have extraordinarily broad definitions of “agricultural land” that make it easy for non-farms to qualify, and state and local governments often do not enforce the few restrictions that do exist or check if former agricultural operations are active.²⁷ Rural property owners can count their land as “farmland” with nominal, and sometimes less than nominal, gestures at agricultural production.

18

any gross income at all from agricultural products, allowing them to collectively deduct $6.8 billion from their returns.³²

As a result of these and other factors, Census of Agriculture data overstate the number of actual farms. At the same time, the data largely ignore many other aspects of the farm economy such as farmworkers, consumers, and residents affected by neighboring farms. This not only distorts economic data such as average income, but it also distorts politics and policymaking. A critical first step in more effective policymaking is the development of a more accurate assessment of actual farm businesses. The Economic Research Service has produced reports on these entities, but the department should make the distinction between business and nonbusiness farms central to its census reports.

We can develop a far more accurate understanding of farm businesses if we use multiple data sources—including the CPS, IRS tax data, and detailed census statistics—rather than relying only on summary agricultural census numbers. In 2017, the USDA census reported more than 2 million “farms” but around 950,000 “farm businesses”³³ and between 1 and 1.4 million operators who said their primary occupation was farming.³⁴ Almost 1.8 million households filed Schedule F forms in 2017, but only about 1 million farms reported gross sales over $50,000. The CPS reported around 900,000 farm households and around 1 million farmers—or people who said their longest job in the previous year was as a farmer—in 2017.³⁵ These three sources all show that there are about half as many farms and farmers as generally reported by USDA and in most press accounts.³⁶

19